Multi-Step Regression Learning for Compositional Distributional Semantics

TL;DR

This paper introduces a novel tensor-based learning method for compositional distributional semantics, demonstrating improved performance on benchmark datasets and potential for addressing complex semantic composition challenges.

Contribution

It presents a new tensor learning approach within a formal semantic framework, outperforming previous methods and offering solutions for more nuanced compositional semantics problems.

Findings

Outperforms existing methods on benchmark datasets

Introduces a generalised tensor learning approach

Potential to solve complex compositional semantics issues

Abstract

We present a model for compositional distributional semantics related to the framework of Coecke et al. (2010), and emulating formal semantics by representing functions as tensors and arguments as vectors. We introduce a new learning method for tensors, generalising the approach of Baroni and Zamparelli (2010). We evaluate it on two benchmark data sets, and find it to outperform existing leading methods. We argue in our analysis that the nature of this learning method also renders it suitable for solving more subtle problems compositional distributional models might face.